July 14, 2010 at 8:40 pm
Does anyone know whether two node active/active clustering is supported with Sql Server 2008? I have found some forum discussions which suggest that it is but I havent seen any documentation from microsoft which says that it is.
July 15, 2010 at 5:47 am
which edition of SQL Server 2008 are you using?
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
July 15, 2010 at 6:57 am
The answer is yes, it does. But, it's important to understand what that means in SQL Server.
What it isn't is a grid clustering solution - e.g. you can not have the same database/SQL Server instance running on two nodes in order to distribute the workload over two servers. In this case, Active/Active means having one SQL Server Instance installed as a Fail-over cluster running on node 1 and another SQL Server Instance installed as a Fail-over cluster running on node 2.
Both sides of the cluster are then Active, but not serving the same Instances.
July 15, 2010 at 8:56 am
also 2 node clusters are only available in Std and Ent editions!
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
July 15, 2010 at 5:59 pm
Thanks for the replies Perry/Howard,
The edition we are using is 2008 Standard Edition. What we want is instance A and instance B on node 1 and instance A and instance B on node 2. In normal running (where failover has not occured), on node 1, instance A will be active and instance B will be inactive. On node 2, instance A will be inactive and instance B will be active. If instance A on node 1 fails then it will failover to node 2 so that now both instance A and instance B are active on node 2. Is this type of setup allowed with SE? We are not attempting to load balance but we would prefer to make some use of both servers rather than having one completely idle as in a Active/Passive configuration.
Also I have been doing some more reading and it looks like a lot of people do not recommend this type of setup due to issues with virtual memory defragmentation upon failover. Is there any nice way to manage this so its not such an issue?
Basically my situation is that we have two existing database servers, each with single Sql Server instances on them, that have no real failover plan. One has 4GB ram (server A) and the other 8GB ram (server B). By putting these two instances into a cluster as explained above I would assume that the optimal node memory setup would be to have 4GB + 8GB = 12GB ram on each server? However we are hoping that we can get away with two 8GB ram servers and if failover occurs then we are prepared to accept the performance issues that will follow.
Does all this sound reasonable?
July 15, 2010 at 7:12 pm
This can work just fine for you as you explained. I would recommend that you try to get as much memory as you can. What kind of Shared storage will you be using?
Also, if you configure your virtual memory correctly, have enough memory you shouldn't need to worry about virtual memory fragmentation (my understanding). That occurs when paging to disk a lot and can happen with or without active active clusters.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
July 15, 2010 at 7:17 pm
Hi Jason,
Thanks for the answer. Will be using SAN for storage.
July 15, 2010 at 7:37 pm
You're welcome.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
July 16, 2010 at 3:54 am
james.christou (7/15/2010)
Thanks for the replies Perry/Howard,The edition we are using is 2008 Standard Edition.
That's fine Std supports a 2 node cluster
james.christou (7/15/2010)
What we want is instance A and instance B on node 1 and instance A and instance B on node 2. In normal running (where failover has not occured), on node 1, instance A will be active and instance B will be inactive. On node 2, instance A will be inactive and instance B will be active. If instance A on node 1 fails then it will failover to node 2 so that now both instance A and instance B are active on node 2. Is this type of setup allowed with SE?
Yes it is!
james.christou (7/15/2010)
We are not attempting to load balance
Good because SQL Server does not do that!
james.christou (7/15/2010)
but we would prefer to make some use of both servers rather than having one completely idle as in a Active/Passive configuration.
perfectly reasonable but you have to configure the cluster correctly or you will have major issues!!
james.christou (7/15/2010)
Also I have been doing some more reading and it looks like a lot of people do not recommend this type of setup due to issues with virtual memory defragmentation upon failover. Is there any nice way to manage this so its not such an issue?
Not an issue if all nodes are the same spec. During fail over the SQL service will stop and databases will shut down, when restarting on the partner node memory will be allocated according to the instance configuration. You are planning to use nodes that have different memory configs.
If inst A has max mem of 2GB and inst B has max mem of 4GB, when they both fail over to node A you will get issues as 2 + 4 = 6 and the node only has 4GB RAM
james.christou (7/15/2010)
Basically my situation is that we have two existing database servers, each with single Sql Server instances on them, that have no real failover plan. One has 4GB ram (server A) and the other 8GB ram (server B). By putting these two instances into a cluster as explained above I would assume that the optimal node memory setup would be to have 4GB + 8GB = 12GB ram on each server? However we are hoping that we can get away with two 8GB ram servers and if failover occurs then we are prepared to accept the performance issues that will follow.Does all this sound reasonable?
Completely wrong my friend that is not how the cluster works. All nodes need to be the same spec, especially RAM
Make sure you understand Windows clustering fully before attempting any configuration work.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
July 17, 2010 at 6:22 pm
Hey Perry, thanks for making that much clearer
July 18, 2010 at 12:50 pm
you're welcome!
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply